Reading the "Haxby" gravity files:

Who is Haxby?

In 1985, Dr. William F. Haxby of the Lamont-Doherty Geological 
Observatory of Columbia University prepared this data base of 
free-air gravity anomalies, based on the radar altimetry of the 
SEASAT satellite mission in 1978.

What about the data?

The utility of SEASAT altimeter data to geophysics is a result of 
their high sensitivity and uniform global coverage. The precision 
of SEASAT measurements is about 0.05m, and the distance between 
adjacent SEASAT tracks is at most 1.5 degrees of longitude. The 
measurement sensitivity was sufficient to detect features of the 
marie gravity field with spatial dimensions as small as 20 km. The 
interpolated gridded data have a latitude and longitude resolution 
of 5 minutes.

The procedure that was used to derive the model of the marine 
gravity field included:  editing the SEASAT measurements to remove 
data spikes; adjusting the profiles to minimize discrepancies 
between measured sea surface heights at intersections between 
SEASAT tracks; estimating the sea surface height at a grid of 
equally spaced pints by fitting least-squares surfaces to the 
edited and adjusted measurements; and, finally, computing marine 
gravity anomalies from the sea surface topography, assuming that 
the surface conforms to the geoid. In order to preserve linear 
trends that are defined by similar sea surface signatures on 
several adjacent SEASAT profiles, lineations that are apparent in 
the along track data were digitized. The digitized lineations 
served as input for an anisotropic surface fitting algorithm that 
was applied in the vicinity of the lineations to revise the 
gridded sea surface topography.

The accuracy of the gravity model is limited by several factors:  
by the accuracy and precision of the measurement; by non-geoidal 
components of the sea surface topography due to currents and 
tides; by gaps between adjacent tracks that are larger than the 
potential resolution of the data; and by smoothing as a result of 
the least-squares gridding procedure. The principal source of 
error in the SEASAT measurements is uncertainty in the satellite's 
position. The application of adjustment procedures to minimize 
crossover differences ensured that these long wavelength errors 
did not cause short wavelength noise, in the form of artificial 
along track lineations, in the gridded gravity model. The profile 
adjustment also reduced some of the inconsistencies introduced by 
temporal oceanographic variability. However, the large amplitudes 
of non-geoidal components associated with ocean dynamics result in 
greater uncertainty in the gravity model in the vicinity of major 
ocean currents and shallow continental shelves. The precision of 
the measurements is not a serious limitation, in light of other 
sources of uncertainty, except at high latitudes where sea ice was 
present. Sea ice is responsible for significantly higher noise 
levels in the region surrounding Antarctica and over Arctic 
continental shelves. The least-squares surface fitting procedure 
resulted in a significantly reduced noise level, but also reduced 
the amplitudes of features of the ocean topography with 
wavelengths less than about 500 km at the equator or 200 km at 
high latitudes.

Format of the data on this disc:

The data represent derived free-air gravity anomalies in 
Milligals. The larger file "HAXBY.ASC" is ASCII characters in 
groups of 4320 five-digit integers per latitude band, 
corresponding to 5-minute steps in longitude at constant latitude 
from longitude 0 D 5' eastward to 360 D 0' longitude, bands 
stepping southward by 5 minutes from latitude 72 Degrees North to 
72 Degrees South. The "record length" is 21600 bytes. The latitude 
for a given record is found by:

     LATITUDE = (865 - RECORD#)/12. (The Equator is record 865.)

Longitude of a point is found by:

     LONGITUDE = (SAMPLE# - 1)/12. (lat and lon in degrees).

The byte offset for a given sample at a specific latitude and 
longitude is given by:

     OFFSET = (72 - LAT)*12*21600 + (LON*12 -1)*5, LAT and LON in 
degrees.

The other data file, "HAXBY.BIN", is organized in the same data 
order, but all data are 16-bit integers in "normal" or hi-byte-
first order (Sun, Macintosh format). The "record length" is 8640 bytes 
per latitude band. To facilitate matching with topographic data
from ETOPO5, the file has been padded with zeros at the north and
south ends to fit from +90D 00' to -89D 55'. Therefore, the offsets
in the BINARY data file are calculated as:

     LATITUDE = (1081 - RECORD#)/12. (The Equator is record 1081.)

Longitude of a point is found by:

     LONGITUDE = (SAMPLE# - 1)/12. (lat and lon in degrees).

The byte offset for a given sample at a specific latitude and 
longitude is given by:

     OFFSET = (90 - LAT)*12*8640 + (LON*12 -1)*2, LAT and LON in 
degrees.

Getting data off the CD-ROM:

Data may be extracted on PCs by the routines "SELSEA1.EXE", "SELSEA5.EXE",
or "SELSEA5B.EXE" in the \GRAVITY\SEASAT directory. The programs UN-SWAP the
data bytes and output PC-ordered bytes or ASCII as needed. You must create
a link to the data file by (in DOS) typing

"SET SEASATD=L:\TOPO\GRAVITY\SEASAT\HAXBY.BIN" (where L: is your CD-ROM drive).

The Macintosh "SELETOPO5" routine may be used to access the data, but you
will have to first copy the "HAXBY.BIN;1" file from the CD-ROM to hard disk.
The first two bytes of the file must be set to 0xEF3E for the program to
recognize the data file. (The image-generation program ETSHADE24 will also 
ecognize the modified file and may be used to generate shaded relief images
from the SEASAT data.) A specific Mac data-extraction program for SEASAT data
may be provided at a later date if interest warrants.

The files "HAXBY.FMT;1" and "HAXBY.HDR;1" are placed in the SEASAT
directory for convenient use of the "FREEFORM" and "GEOVU2"
utilities being developed at NGDC. (These programs were not available
for inclusion on the CD -- watch for an announcement of their release.)